A Two-Ball Esllberg Paradox: Experimental Evidence

We conduct an incentivized experiment on a nationally representative US sample (N=708) to test how people prefer to avoid ambiguity even when the ambiguity improves the probability of receiving a fixed price. We find that subjects prefer non-ambiguous acts to similar ambiguous acts, even when the ambiguous acts provide larger win probabilities. Furthermore, this preference for avoiding ambiguity is not entirely due to a lack of understanding, as subjects "correctly" select the act with a larger win probability when comparing two similar ambiguous acts. Traditional models of ambiguity aversion cannot explain such preferences.